@tychenjiajun/pi-voice 0.2.0 → 0.4.0

package/src/settings-menu.ts

@@ -2,11 +2,10 @@ import type {
   ExtensionAPI,
   ExtensionContext,
 } from "@earendil-works/pi-coding-agent";
-import { Box, Container, Input, Spacer, Text, truncateToWidth } from "@earendil-works/pi-tui";
+import { truncateToWidth } from "@earendil-works/pi-tui";
 import { microphonePermission, type Permission } from "./audio.js";
 import { displayLanguage, getCatalogModel } from "./catalog.js";
 import { chineseOutputSummary, isChineseLanguage } from "./chinese.js";
-import { VoiceKeys } from "./keybindings.js";
 import {
   chooseLanguages,
   createTranscriptionLanguagePicker,
@@ -18,20 +17,20 @@ import {
   microphoneSummary,
 } from "./microphone-picker.js";
 import { runModelSelection } from "./onboarding.js";
+import { chooseBackendType, promptRemoteBackend, warnMissingApiKey } from "./remote-asr.js";
 import {
   writeSettings,
   type BackendSettings,
   type ChineseOutput,
   type MicrophoneSetting,
+  type PolishModelRef,
   type TranscribeSettings,
   type TranscriptionLanguage,
 } from "./settings.js";
 import { displayShortcut } from "./shortcut-core.js";
 import { createShortcutPicker } from "./shortcuts.js";
 import {
-  LIST_PADDING,
   padToWidth,
-  panelBorder,
   SingleSelectPicker,
   type SingleSelectChoice,
 } from "./ui-components.js";
@@ -42,16 +41,17 @@ const SETTINGS_LABEL_WIDTH = 25;
 // Text padding (2), cursor gutter (2), label column, and its gap (2).
 const SETTINGS_ROW_OVERHEAD = SETTINGS_LABEL_WIDTH + 6;
 
-type SettingsAction =
+export type SettingsAction =
   | "preferred-languages"
   | "model"
   | "backend"
   | "transcription-language"
   | "chinese-output"
+  | "post-processing"
   | "microphone"
   | "shortcut";
 
-type SettingsHomeChoice = SingleSelectChoice<SettingsAction> & {
+export type SettingsHomeChoice = SingleSelectChoice<SettingsAction> & {
   summary: string;
   /** Render the summary in the error color. */
   alert?: boolean;
@@ -91,7 +91,12 @@ function backendSummary(backend: BackendSettings): string {
   return backend.type === "openai-asr" ? `Remote · ${backend.model}` : "Local model";
 }
 
-function settingsHomeChoices(
+function polishSummary(polishEnabled: boolean | undefined, model: PolishModelRef | undefined): string {
+  if (!polishEnabled) return "Off";
+  return model ? `${model.provider}/${model.id}` : "Pi's current model";
+}
+
+export function settingsHomeChoices(
   configured: TranscribeSettings,
   permission: Permission,
 ): SettingsHomeChoice[] {
@@ -116,17 +121,34 @@ function settingsHomeChoices(
       description:
         "Local model, or an OpenAI-compatible remote ASR endpoint (local model stays as fallback)",
     },
+    {
+      value: "post-processing",
+      label: "Post-processing",
+      summary: polishSummary(configured.polishEnabled, configured.polishModel),
+      description:
+        "Refine transcripts with Pi's current model after ASR, like context compaction",
+    },
     {
       value: "transcription-language",
       label: "Transcription language",
       summary: transcriptionLanguageSummary(configured.transcriptionLanguage, model),
-      description: "Language expected in recordings, or automatic detection when supported",
+      description:
+        configured.backend.type === "openai-asr"
+          ? "Used by the local fallback and Chinese output conversion; the remote endpoint detects the language itself"
+          : "Language expected in recordings, or automatic detection when supported",
     },
   ];
 
+  // Chinese script conversion applies when the configured language is Chinese,
+  // when a preferred language is Chinese, or when a detecting model can pick
+  // Chinese on its own. Gating only on the first two leaves the row unreachable
+  // for anyone relying on auto-detection.
+  const detectsChinese =
+    model.capabilities.languageDetection && model.languages.some(isChineseLanguage);
   if (
     isChineseLanguage(configured.transcriptionLanguage) ||
-    configured.preferredLanguages.some(isChineseLanguage)
+    configured.preferredLanguages.some(isChineseLanguage) ||
+    detectsChinese
   ) {
     choices.push({
       value: "chinese-output",
@@ -266,212 +288,158 @@ async function chooseShortcut(
   );
 }
 
-/**
- * Single-field text prompt built on pi-tui's Input. Empty input may be legal
- * (the caller decides via validate); Escape backs out of the whole flow.
- */
-function promptText(
+/** Backend row flow: pick the type, then collect remote settings step by step. */
+async function editBackend(
   ctx: ExtensionContext,
-  options: {
-    title: string;
-    description?: string;
-    value: string;
-    placeholder?: string;
-    validate?: (value: string) => string | undefined;
-  },
-): Promise<string | undefined> {
-  return ctx.ui.custom<string | undefined>((tui, theme, keybindings, done) => {
-    const keys = new VoiceKeys(keybindings);
-    const input = new Input();
-    input.setValue(options.value);
-    let error: string | undefined;
-    const container = new Container();
-
-    const rebuild = (): void => {
-      container.clear();
-      container.addChild(panelBorder(theme));
-      container.addChild(new Spacer(1));
-      container.addChild(new Text(theme.fg("accent", theme.bold(options.title)), 1, 0));
-      if (options.description) {
-        container.addChild(new Text(theme.fg("muted", options.description), 1, 0));
-      }
-      container.addChild(new Spacer(1));
-      if (error) {
-        container.addChild(new Text(theme.fg("error", error), 1, 0));
-        container.addChild(new Spacer(1));
-      }
-      if (!input.getValue() && options.placeholder) {
-        container.addChild(new Text(theme.fg("dim", options.placeholder), 2, 0));
-      }
-      const inputBox = new Box(LIST_PADDING, 0);
-      inputBox.addChild(input);
-      container.addChild(inputBox);
-      container.addChild(new Spacer(1));
-      container.addChild(
-        new Text(
-          `${keys.hint("tui.input.submit", "save")}  ${keys.hint("tui.select.cancel", "back")}`,
-          1,
-          0,
-        ),
-      );
-      container.addChild(new Spacer(1));
-      container.addChild(panelBorder(theme));
-    };
-    rebuild();
-
-    input.onSubmit = (value) => {
-      const problem = options.validate?.(value.trim());
-      if (problem) {
-        error = problem;
-        rebuild();
-        tui.requestRender();
-        return;
-      }
-      done(value.trim());
-    };
-    input.onEscape = () => done(undefined);
+  configured: TranscribeSettings,
+): Promise<boolean> {
+  const type = await chooseBackendType(ctx, configured.backend);
+  if (!type) return false;
+  if (type === "transcribe-cpp") {
+    if (configured.backend.type === "transcribe-cpp") return false;
+    return saveUpdatedSettings(
+      ctx,
+      configured,
+      { ...configured, backend: { type: "transcribe-cpp" } },
+      "Using the local model",
+    );
+  }
 
-    return {
-      get focused() {
-        return input.focused;
-      },
-      set focused(value: boolean) {
-        input.focused = value;
-      },
-      render(width: number): string[] {
-        return container
-          .render(width)
-          .map((line) => truncateToWidth(line, width, ""));
-      },
-      handleInput(data: string): void {
-        input.handleInput(data);
-      },
-      invalidate(): void {
-        container.invalidate();
-      },
-    };
-  });
+  const current =
+    configured.backend.type === "openai-asr" ? configured.backend : undefined;
+  const backend = await promptRemoteBackend(ctx, current);
+  if (!backend) return false;
+
+  const saved = await saveUpdatedSettings(
+    ctx,
+    configured,
+    { ...configured, backend },
+    `Remote ASR configured (${backend.model})`,
+  );
+  if (saved) warnMissingApiKey(ctx, backend.apiKeyEnv);
+  return saved;
 }
 
-async function chooseBackendType(
+async function choosePolishEnabled(
   ctx: ExtensionContext,
-  current: BackendSettings,
-): Promise<"transcribe-cpp" | "openai-asr" | undefined> {
-  const choices: SingleSelectChoice<"transcribe-cpp" | "openai-asr">[] = [
+  current: boolean,
+): Promise<boolean | undefined> {
+  const choices: SingleSelectChoice<"off" | "on">[] = [
     {
-      value: "transcribe-cpp",
-      label: "Local model",
-      description: "Transcribe fully on-device with the downloaded model; works offline",
+      value: "off",
+      label: "Off",
+      description: "Insert transcripts exactly as ASR output them",
     },
     {
-      value: "openai-asr",
-      label: "Remote ASR (OpenAI-compatible)",
+      value: "on",
+      label: "On",
       description:
-        "Upload audio to any OpenAI-compatible chat-completions ASR endpoint; falls back to the local model if the request fails",
+        "Refine transcripts after ASR with a Pi model and its provider auth, like context compaction; keeps the raw transcript on failure",
     },
   ];
-  return ctx.ui.custom<"transcribe-cpp" | "openai-asr" | undefined>(
+  const chosen = await ctx.ui.custom<"off" | "on" | undefined>(
     (tui, theme, keybindings, done) =>
       new SingleSelectPicker(
         tui,
         theme,
         keybindings,
         choices,
-        current.type,
-        { title: "Transcription backend", cancelLabel: "back" },
+        current ? "on" : "off",
+        { title: "Post-processing", cancelLabel: "back" },
         done,
       ),
   );
+  return chosen === undefined ? undefined : chosen === "on";
 }
 
-function validateBaseUrl(value: string): string | undefined {
-  if (!value) return "Base URL is required";
-  try {
-    const url = new URL(value);
-    if (url.protocol !== "http:" && url.protocol !== "https:") {
-      return "Base URL must start with http:// or https://";
-    }
-  } catch {
-    return "Base URL must be a valid URL";
-  }
-  return undefined;
+function polishModelKey(model: PolishModelRef): string {
+  return `${model.provider}/${model.id}`;
+}
+
+function samePolishModel(
+  left: PolishModelRef | undefined,
+  right: PolishModelRef | undefined,
+): boolean {
+  if (!left || !right) return left === right;
+  return left.provider === right.provider && left.id === right.id;
 }
 
-function validateEnvName(value: string): string | undefined {
-  if (!value) return "Environment variable name is required";
-  if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(value)) {
-    return "Use a plain environment variable name, e.g. OPENAI_API_KEY";
+/** Pick Pi's current model (default) or any model with configured auth. */
+async function choosePolishModel(
+  ctx: ExtensionContext,
+  current: PolishModelRef | undefined,
+): Promise<"current" | PolishModelRef | undefined> {
+  const available = ctx.modelRegistry.getAvailable();
+  const models = available.length > 0 ? available : ctx.modelRegistry.getAll();
+  const choices: SingleSelectChoice<string>[] = [
+    {
+      value: "current",
+      label: "Pi's current model",
+      description: "Use whichever model the agent is using (default)",
+    },
+  ];
+  const seen = new Set<string>();
+  for (const model of models) {
+    const key = `${model.provider}/${model.id}`;
+    if (key === "current" || seen.has(key)) continue;
+    seen.add(key);
+    choices.push({ value: key, label: key, description: model.name });
   }
-  return undefined;
+  if (current && !seen.has(polishModelKey(current))) {
+    choices.push({
+      value: polishModelKey(current),
+      label: polishModelKey(current),
+      description: "Configured post-processing model (not currently available)",
+    });
+  }
+  const chosen = await ctx.ui.custom<string | undefined>((tui, theme, keybindings, done) =>
+    new SingleSelectPicker(
+      tui,
+      theme,
+      keybindings,
+      choices,
+      current ? polishModelKey(current) : "current",
+      { title: "Post-processing model", cancelLabel: "back" },
+      done,
+    ),
+  );
+  if (chosen === undefined) return undefined;
+  if (chosen === "current") return "current";
+  const slash = chosen.indexOf("/");
+  return { provider: chosen.slice(0, slash), id: chosen.slice(slash + 1) };
 }
 
-/** Backend row flow: pick the type, then collect remote settings step by step. */
-async function editBackend(
+/** Post-processing row flow: toggle, then choose Pi's current model or an override. */
+async function editPostProcessing(
   ctx: ExtensionContext,
   configured: TranscribeSettings,
 ): Promise<boolean> {
-  const type = await chooseBackendType(ctx, configured.backend);
-  if (!type) return false;
-  if (type === "transcribe-cpp") {
-    if (configured.backend.type === "transcribe-cpp") return false;
+  const enabled = await choosePolishEnabled(ctx, configured.polishEnabled ?? false);
+  if (enabled === undefined) return false;
+  if (!enabled) {
+    if (!configured.polishEnabled) return false;
     return saveUpdatedSettings(
       ctx,
       configured,
-      { ...configured, backend: { type: "transcribe-cpp" } },
-      "Using the local model",
+      { ...configured, polishEnabled: false },
+      "Post-processing disabled",
     );
   }
 
-  const current =
-    configured.backend.type === "openai-asr" ? configured.backend : undefined;
-  const baseUrl = await promptText(ctx, {
-    title: "ASR base URL",
-    description: "OpenAI-compatible endpoint root, without the /chat/completions path",
-    value: current?.baseUrl ?? "",
-    placeholder: "https://api.example.com/v1",
-    validate: validateBaseUrl,
-  });
-  if (baseUrl === undefined) return false;
-
-  const model = await promptText(ctx, {
-    title: "ASR model",
-    value: current?.model ?? "",
-    placeholder: "whisper-1",
-    validate: (value) => (value ? undefined : "Model name is required"),
-  });
-  if (model === undefined) return false;
-
-  const apiKeyEnv = await promptText(ctx, {
-    title: "API key environment variable",
-    description: "The key itself is read from this variable at runtime; it is never stored in settings",
-    value: current?.apiKeyEnv ?? "",
-    placeholder: "OPENAI_API_KEY",
-    validate: validateEnvName,
-  });
-  if (apiKeyEnv === undefined) return false;
-
-  const backend: BackendSettings = { type: "openai-asr", baseUrl, model, apiKeyEnv };
-  if (
-    current &&
-    current.baseUrl === baseUrl &&
-    current.model === model &&
-    current.apiKeyEnv === apiKeyEnv
-  ) {
+  const selection = await choosePolishModel(ctx, configured.polishModel);
+  if (selection === undefined) return false;
+  const polishModel = selection === "current" ? undefined : selection;
+  if (configured.polishEnabled && samePolishModel(configured.polishModel, polishModel)) {
     return false;
   }
-  const saved = await saveUpdatedSettings(
+  return saveUpdatedSettings(
     ctx,
     configured,
-    { ...configured, backend },
-    `Remote ASR configured (${model})`,
+    { ...configured, polishEnabled: true, polishModel },
+    polishModel
+      ? `Post-processing enabled (${polishModelKey(polishModel)})`
+      : "Post-processing enabled (Pi's current model)",
   );
-  if (saved && !process.env[apiKeyEnv]) {
-    ctx.ui.notify(
-      `${apiKeyEnv} is not set in Pi's environment; remote transcription will fall back to the local model until it is set`,
-      "warning",
-    );
-  }
-  return saved;
 }
 
 export async function openMacOSMicrophoneSettings(
@@ -549,6 +517,8 @@ export async function showSettingsMenu(
         currentModelId: configured.model.id,
         microphone: configured.microphone,
         backend: configured.backend,
+        polishEnabled: configured.polishEnabled,
+        polishModel: configured.polishModel,
         postActivation: "stay",
         onPreferredLanguagesChange: async (preferredLanguages) => {
           if (languagesEqual(preferredLanguages, configured.preferredLanguages)) return;
@@ -567,6 +537,11 @@ export async function showSettingsMenu(
       continue;
     }
 
+    if (action === "post-processing") {
+      await editPostProcessing(ctx, configured);
+      continue;
+    }
+
     if (action === "transcription-language") {
       const transcriptionLanguage = await chooseTranscriptionLanguage(ctx, configured);
       if (

package/src/settings.ts

@@ -28,6 +28,12 @@ export type BackendSettings =
   | { type: "transcribe-cpp" }
   | { type: "openai-asr"; baseUrl: string; model: string; apiKeyEnv: string };
 
+/**
+ * Explicit model for post-processing, resolved through Pi's model registry and
+ * its provider auth. Absent means the agent's current model.
+ */
+export type PolishModelRef = { provider: string; id: string };
+
 function defaultChineseOutput(): ChineseOutput {
   const locale = Intl.DateTimeFormat().resolvedOptions().locale;
   const subtags = locale.toLowerCase().split("-");
@@ -47,6 +53,10 @@ export type TranscribeSettings = {
   transcriptionLanguage: TranscriptionLanguage;
   chineseOutput: ChineseOutput;
   microphone: MicrophoneSetting;
+  /** Optional post-processing with Pi's current model; absent means disabled (backward compatible). */
+  polishEnabled?: boolean;
+  /** Optional post-processing model override; absent uses the agent's current model. */
+  polishModel?: PolishModelRef;
   model: {
     source: "catalog";
     id: string;
@@ -147,6 +157,13 @@ function validateBackend(value: unknown): BackendSettings | undefined {
   return undefined;
 }
 
+function normalizePolishModel(value: unknown): PolishModelRef | undefined {
+  if (!isObject(value)) return undefined;
+  if (typeof value.provider !== "string" || !value.provider.trim()) return undefined;
+  if (typeof value.id !== "string" || !value.id.trim()) return undefined;
+  return { provider: value.provider.trim(), id: value.id.trim() };
+}
+
 function validateSettings(value: unknown): TranscribeSettings | undefined {
   if (!isObject(value) || value.version !== SETTINGS_VERSION) return undefined;
   const backend = validateBackend(value.backend);
@@ -163,6 +180,7 @@ function validateSettings(value: unknown): TranscribeSettings | undefined {
   const preferredLanguages = normalizeLanguages(value.preferredLanguages);
   const microphone = validateMicrophone(value.microphone);
   if (!preferredLanguages || !microphone) return undefined;
+  const polishModel = normalizePolishModel(value.polishModel);
 
   return {
     version: SETTINGS_VERSION,
@@ -176,6 +194,8 @@ function validateSettings(value: unknown): TranscribeSettings | undefined {
     ),
     chineseOutput: validateChineseOutput(value.chineseOutput),
     microphone,
+    polishEnabled: value.polishEnabled === true,
+    ...(polishModel ? { polishModel } : {}),
     model: {
       source: "catalog",
       id: value.model.id,
@@ -244,6 +264,10 @@ type ModelSettingsOptions = {
   microphone?: MicrophoneSetting;
   /** Preserved across model re-selection; defaults to the local backend. */
   backend?: BackendSettings;
+  /** Preserved across model re-selection; defaults to disabled. */
+  polishEnabled?: boolean;
+  /** Preserved across model re-selection; absent means the agent's current model. */
+  polishModel?: PolishModelRef;
 };
 
 export function settingsForModel(
@@ -268,6 +292,8 @@ export function settingsForModel(
     ),
     chineseOutput: options.chineseOutput ?? defaultChineseOutput(),
     microphone: { ...(options.microphone ?? DEFAULT_MICROPHONE) },
+    polishEnabled: options.polishEnabled ?? false,
+    ...(options.polishModel ? { polishModel: options.polishModel } : {}),
     model: { source: "catalog", id: modelId, path: modelPath },
   };
 }

package/src/try-it.ts

@@ -189,7 +189,7 @@ export class TryItPane implements Component {
     const topChrome = (compact: boolean): string[] => [
       ...panelBorder(this.theme).render(width),
       ...(compact ? [] : [""]),
-      ...onboardingHeader(this.theme, title, 3).render(width),
+      ...onboardingHeader(this.theme, title, 4).render(width),
       ...(compact ? [] : [""]),
       ...(compact ? [line(instructions(true))] : text(instructions(false))),
       ...(compact ? [] : [""]),
@@ -208,7 +208,12 @@ export class TryItPane implements Component {
         ...permission,
         ...render(setting("Shortcut", shortcut, this.keys.keyText("voice.tryIt.shortcut"), compact)),
         ...render(setting("Microphone", microphoneSummary(this.settings.microphone), this.keys.keyText("voice.tryIt.microphone"), compact)),
-        ...render(setting("Model", modelName, this.keys.keyText("voice.tryIt.model"), compact)),
+        ...render(setting(
+          this.settings.backend.type === "openai-asr" ? "Fallback" : "Model",
+          modelName,
+          this.keys.keyText("voice.tryIt.model"),
+          compact,
+        )),
         ...(compact ? [] : [""]),
         ...text(hints),
         ...(compact ? [] : [""]),
@@ -230,7 +235,7 @@ export class TryItPane implements Component {
     if (top.length + bottom.length + frameRows + 1 > budget) {
       // Tiny terminals: drop settings, but keep the task and current actions visible.
       top = [
-        ...onboardingHeader(this.theme, title, 3).render(width),
+        ...onboardingHeader(this.theme, title, 4).render(width),
         line(instructions(true)),
       ];
       bottom = [...text(hints), ...panelBorder(this.theme).render(width)];

package/src/ui-components.ts

@@ -37,7 +37,7 @@ export function onboardingHeader(
   theme: UiTheme,
   title: string,
   step: number,
-  total = 3,
+  total = 4,
 ): Component {
   const context = `Pi Voice setup · ${step} of ${total}`;
   const compactContext = `${step} of ${total}`;
@@ -217,6 +217,8 @@ export class SingleSelectPicker<T extends string> extends Container implements F
       searchable?: boolean;
       maximumVisible?: number;
       cancelLabel?: string;
+      /** Adds setup context and progress to the title. */
+      onboardingStep?: number;
       /** Extra footer legend, appended after the ✓ current marker. */
       legend?: string;
       /** Custom row body after the cursor and ✓ markers; handles its own active styling. */
@@ -245,7 +247,11 @@ export class SingleSelectPicker<T extends string> extends Container implements F
 
     this.addChild(panelBorder(theme));
     this.addChild(new Spacer(1));
-    this.addChild(this.titleText);
+    if (typeof options.onboardingStep === "number") {
+      this.addChild(onboardingHeader(theme, options.title, options.onboardingStep));
+    } else {
+      this.addChild(this.titleText);
+    }
     if (this.subtitleText) this.addChild(this.subtitleText);
     this.addChild(new Spacer(1));
     if (options.searchable) {